150 ALASKAN GLACIER STUDIES 40 feet apart, and with a depth of 5 to 10 feet. There is every reason for believing that these channelways had been occupied by water derived from the melting of the snows in the spring of 1905. Further proof that the surface of almost all parts of the gravel plain was frequently reached by the branching glacial streams is found in the absence of vegetation, for even individual annual plants were found in only a few places. With such rapid upbuilding of the plain as was evidently in progress, and with such recent extension of the streams over its surface, it is inconceivable that kettles could long remain, for they would soon be destroyed by stream erosion and by gravel deposition. We are convinced, therefore, that the kettles observed in 1905 were in the main the result of the melting of the buried ice in that single season. The presence of buried ice beneath the outwash gravel plain, and its direct connection with Hidden Glacier, are established by the facts observed in 1905. We could not determine exactly how far the ice extended, though the presence of kettles to a distance of at least a mile from the ice front furnished evidence of its extension that far. Beyond this point no kettles were observed, and while it is possible that ice may have existed further than a mile from the visible ice front, it is not probable, and, if it did, it was certainly very thin. The inference that we draw from the facts observed, is that a moderately-thick ice foot, buried beneath the gravels, projected for approximately a mile beyond the visible front of Hidden Glacier, gradually thinning and disappearing. There was a continuous diminution in the size of kettles and in the degree of slumping of the surface for about a mile from the ice front. This of course may in part have been due to deeper burial of the ice on the outer portion, but it was probably mainly due to the thinning of the ice. It is a noteworthy fact that the area of most abundant and largest kettles, which was within a quarter of a mile of the visible glacier front, was on the site occupied by the glacier when Gilbert photographed it in 1899. The presence of this buried ice beneath the gravel plain, and the evidence of its direct connection with the glacier beneath the fosse, supports the conclusion reached above that Hidden Glacier was not in active movement in 1905, at least in its outermost portion. It would have required only a very slight thrust to have broken the buried ice, and with it the overlying gravel plain. Even at the very edge of the fosse the gravel showed no sign whatsoever of faulting, with the exception of that due to subsidence. We infer from this fact, as well ad from the uncrevassed glacier surface and the evidence of recession of its visible front, that in 1905 the lower part of Hidden Glacier was in a stagnant state, and that its recession would in all probability have continued, as in the past years, were it not for the absolute change in conditions brought about by the earthquakes in 1899. The manner in which we conceive the burial of the glacier terminus by the gravels to have been effected is as follows: In the early spring the expanded surface of Hidden Glacier reached at least as far out as the west wall of the fosse, and probably farther. With the coming of spring, large streams, derived from melting of snow upon the glacier deposited gravel upon the glacier terminus, thus protecting it from melting rapidly. When such streams waned and ceased depositing, ablation would proceed more rapidly upon the clear ice portion of the glacier than upon that veneered with gravel. In this way a fosse or valley developed between the clear and the gravel-veneered ice. A lake might form in such a depression, ultimately being drained through moulins. Where the gravel was thinnest, that is nearest the glacier, the subsequent, more rapid melting of the ice beneath the gravels resulted in slumping, with the production of a kame moraine.